[Eeglablist] When to perform re-reference?

Fri Jun 27 22:11:01 PDT 2014

It is part of my discussion with Jason. Hope it is helpful:

Hi Iman,

When you do re-referencing, you impose a condition on the estimated map vectors. E.g. average reference enforces the sum over the channels of the map potential to equal zero (that the map be orthogonal to e, the vector of all 1’s). Average mastoid reference will guarantee that the average of the mastoid channels in all the learned maps will be zero. Basically you make all the learned maps orthogonal to a certain direction, and you lose a degree of freedom, or a dimension in the rank of the data.

If the head had perfect sensor coverage over the head “sphere”, and there was limited shunting of electric field due to bone and tissue, then a source inside the head would be expected to sum to zero by charge conservation—any instantaneous increase in charge in one area must be accompanied by decrease in the charge in another area.

However, e.g. in the case of a radial source on the cortical surface, the negative field detected by electrodes on the side of the head/neck opposite the radial source may not have sufficient coverage or may be shunted to the extent that the source should appears to have a net positive charge, with weaker negative field detected relative to positive field. So the “actual” net charge may be non-zero.

Average reference though enforces zero net charge, and thus will in effect add a negative constant to each channel, decreasing the positivity, increasing the negativity, resulting in a radial dipolar source that appear to be deeper in the head than it actually is.

In dipole fitting, a number of factors are taken into consideration, including residual variance of a dipolar model, and physiological plausibility given say a known MRI. I would suggest that dipole fitting of a learned ICA map be carried out by optimizing the fit and plausibility of the source over both (1) dipole location/orientation, and (2) unknown constant channel offset, particularly for radial sources.

I believe that radial sources are often localized to deeper than MRI cortical surface, which is consistent with this idea. However, given the imprecision in tissue conductance values (see Akalin-Acar and Makeig 2013), and the general controversy over the nature of EEG sources themselves (are they confined to cortical patches, or do they represent a larger scale charge redistribution involving sub-cortical and cortical regions?), it is difficult to be certain of anything.

The main point about ICA being reference-free is that it however you reference the data, it won’t limit the ability of ICA to find the source. You are just imposing the reference condition on the estimated map and losing a degree of freedom (location along a single dimension). If you presume that radial sources are confined to cortical patches, then the depth ambiguity can be resolved using prior MRI information.

Note: this should not be taken as an official position of any organizations or communities, just my own comments argued such as they are.

Best,

Jason

===============================

From: eeglablist-bounces at sccn.ucsd.edu [mailto:eeglablist-bounces at sccn.ucsd.edu] On Behalf Of Makoto Miyakoshi
Sent: Friday, June 27, 2014 9:46 PM
To: Jerry Zhu; Jason Palmer
Cc: EEGLAB List
Subject: Re: [Eeglablist] When to perform re-reference?

Dear Jerry,

> Well, this wiki page says average referencing before ICA could reduce the rank by 1, which seems to be a bad thing. 

Yeah, I thought so too. But I hear Jason Palmer, who wrote AMICA, say data should be average referenced before ICA... I add Jason here again, expecting to hear from him on this issue.

Makoto

On Thu, Jun 26, 2014 at 1:55 PM, Jerry Zhu <jerryzhu at siu.edu <mailto:jerryzhu at siu.edu> > wrote:

Well, this wiki page says average referencing before ICA could reduce the rank by 1, which seems to be a bad thing. 

So I guess, average reference should go after ICA.

http://sccn.ucsd.edu/wiki/Chapter_09:_Decomposing_Data_Using_ICA

close to the end of page
"When computing average reference on n-channel data, the rank of the data is reduced to n-1. Why? Because the sum of the potential is 0 at all time points, the last channel activity is equal to minus the sum of the others. ICA does not behave well in this (rank-deficient) condition."

--
Jian Zhu, M.A.
Brain and Cognitive Sciences
Department of Psychology
Southern Illinois University Carbondale

Web: http://zhupsy.com

We have two halves in the brain: left and right. Nothing is right in the left. Nothing is left in the right.

On Mon, Jun 23, 2014 at 11:56 AM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu <mailto:mmiyakoshi at ucsd.edu> > wrote:

Dear Brian,

The average referencing is not affected by epoch rejection (assuming that bad EEG is restricted to the epochs that are rejected) or component rejection. In your case, average referencing can be done either before or after two ICAs.

I'm still waiting to hear from Jason though whether it's better to do average referencing before ICA and if so, why.

Makoto

On Thu, Jun 19, 2014 at 8:37 AM, Brian Scally <scallybrian at gmail.com <mailto:scallybrian at gmail.com> > wrote:

Hi there,

I have been following the discussion. So it is recommended to re-reference to average after removing channels/epochs and right before running ICA. Suppose I plan on running ICA twice: the first time to identify epochs that are affected by improbable components, after which the epochs will be removed. The second time would be to remove actual components. Do I need to re-reference before the first, or will before the second suffice?

Cheers,

Brian

On 19 June 2014 00:56, Makoto Miyakoshi <mmiyakoshi at ucsd.edu <mailto:mmiyakoshi at ucsd.edu> > wrote:

yes.

On Wed, Jun 18, 2014 at 3:45 PM, Jerry Zhu <jerryzhu at siu.edu <mailto:jerryzhu at siu.edu> > wrote:

Hi Makoto,

Thank you again for your reply.

Did you mean "averaging AFTER removing bad epoches" in your sentence "you should apply average referencing AFTER removing bad channels."?

So here are my processing:

1) re-reference  (from Cz to average in my case)

2) remove bad epoches

3) interpolate bad channels

4) ICA

Sounds like you would recommend the order: 2143?

Looking forward to your and others inputs!

Thanks all!
Jerry

--
Jian Zhu, M.A.
Brain and Cognitive Sciences
Department of Psychology
Southern Illinois University Carbondale

Web: http://zhupsy.com

We have two halves in the brain: left and right. Nothing is right in the left. Nothing is left in the right.

On Wed, Jun 18, 2014 at 1:38 PM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu <mailto:mmiyakoshi at ucsd.edu> > wrote:

Dear Jason,

I remember you said that average reference should be applied before ICA, but I did not understand exactly why. Could you comment on this?

Jerry, you should apply average referencing AFTER removing bad channels.

I don't think it makes big differences between average referencing before or after interpolation. However we don't necessarily recommend interpolation before ICA.

Makoto

On Tue, Jun 17, 2014 at 12:20 PM, Jerry Zhu <jerryzhu at siu.edu <mailto:jerryzhu at siu.edu> > wrote:

Hi all,

At which stage do you re-reference your data? It was suggested do re-ref before ICA (http://sccn.ucsd.edu/pipermail/eeglablist/2011/003795.html). How about re-ref before/after bad epochs rejection and bad channel interpolation? The note here  (ftp://ftp.egi.com/pub/documentation/technotes/SplineInterpolation.pdf) suggests after interpolation. ("Since the interpolated potentials can be used to better approximate the average reference (Junghofer et al., 1999), it may be somewhat advantageous to compute the interpolated potentials directly from the measured data, in which case the voltages below are best thought of as referring to these measured potentials")

What is your experience? Thanks for your sharing and suggestions!

Jerry

--
Jian Zhu, M.A.
Brain and Cognitive Sciences
Department of Psychology
Southern Illinois University Carbondale

Web: http://zhupsy.com

We have two halves in the brain: left and right. Nothing is right in the left. Nothing is left in the right.

_______________________________________________
Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu <mailto:eeglablist-unsubscribe at sccn.ucsd.edu> 
For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu <mailto:eeglablist-request at sccn.ucsd.edu> 

-- 

Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego

-- 

Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego

_______________________________________________
Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu <mailto:eeglablist-unsubscribe at sccn.ucsd.edu> 
For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu <mailto:eeglablist-request at sccn.ucsd.edu> 

-- 

Brian Scally 

-- 

Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego

-- 

Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20140627/d14ffed5/attachment.html>